Cross-View User Identification

Vision-aided systems must identify and track multiple communication users among distractors in crowded environments to enable precise beam\/resource allocation for each target user. In this regard, a user identification method for UAV-to-vehicle networks utilizes multimodal sensor data (vehicle compass, UAV\/vehicle-view images) fed into a non-shared-weights Siamese neural network to identify the target vehicle among non-transmitting\/non-receiving ones. This dataset is used to train this model for bridging the gap between ground and aerial views, thereby matching radio and visual domains.